The Science Behind Kinedu

The Science behind Kinedu

Most parenting advice is opinion. Kinedu is built on evidence — millions of real developmental observations, a research partnership with Stanford, and decades of child-development science — distilled into a daily plan made for your child.

Stanford
MIT
The research question

What does early development actually look like — not in a lab, but in millions of ordinary homes?

“The broad sequence is shared.
The calendar is personal.”

Before any statistic — the product itself

Every data point starts as a question a parent can answer.

No clinic, no equipment — just what you see in ordinary moments. This is a real question from the 12-month assessment:

Physical · Standing up

“Can stand up from a squat all alone.”

Not yet Yes

49.5% of 12-month-olds show it · asked right at the edge, where an answer carries information · Example — not a real child's result.

One continuous story

Built by Kinedu. Studied by Stanford.
Tested at population scale.

An instrument made with caregivers, an independent developmental study, and expected ages tested against the full Kinedu record.

1
2,135
caregivers

Kinedu created the caregiver-report instrument and tested its initial psychometric properties with 2,135 Spanish-speaking caregivers. Michael C. Frank, of Stanford, advised on the early revision of its 0–24-month milestone items.

2
21,861
children

Stanford researchers Stenhaug, Ram and Frank analyzed 21,861 children and 177,934 observations from Kinedu data in The Structure of Developmental Variation in Early Childhood.

3
3.5M
children

Kinedu compared its expected ages against approximately 280.6 million in-scope milestone answers from 3.5 million children.

Each figure with its exact unit — caregivers, children and answers are not interchangeable, and we keep them straight.

How the instrument thinks

It doesn't ask what your child already does.
It asks right at the edge.

A question every child passes teaches us nothing about yours. The informative questions are the ones about half of children pass at that age — which means for a perfectly typical child, many honest answers are “not yet”. That's the instrument working, not your child falling short.

AgeExpected “yes”Expected “not yet”
3 months11 / 3221 / 32
12 months23 / 329 / 32
24 months26 / 326 / 32

Expected answers for a typical child. “Not yet” is as informative as “yes”.

100% 75% 50% 25% 0% 17% 48% 93% 0 months 3 months 5 months

“Lifts head while on tummy” — share of babies showing it, by age · Kinedu record, 2020–2023. This curve is what turns a “yes” into information.

Why 32 questions

Reliability 0.92 — right where the curve flattens.

Eight questions per area reach 0.87. Thirty-two reach 0.92. Answering 48 barely adds anything — so we stop where your time stops paying for precision. The free reading is the most reliable number the product produces.

1.00 0.75 0.50 0.25 0 8 · 0.87 16 · 0.90 32 · 0.92 48 · 0.93 the free reading ↑

Reliability of the overall reading by number of questions answered.

What we found

Four findings, each with its number — and its limit.

From Coupling and Differentiation in Early Development, Kinedu's population-scale replication and psychometric characterisation (2026).

Expected ages match reality. Every milestone's expected age was checked against ~280 million in-scope observations.

median error: 0 months vs the empirical 75th percentile

In year one, development moves as one connected system. First described by Stanford researchers in Kinedu data; reproduced by Kinedu across ~1.7 million children.

within-child coupling across areas, replicated at 2-, 4- and 6-month intervals

Around age one, the areas begin keeping their own time. Coupling softens — it does not vanish.

connection roughly halves at the transition: 0.49 → 0.27

A starting point is not a destiny. Where a child starts helps explain where they stand later — and says almost nothing about how fast they move next.

37% of later standing · 1.5% of subsequent pace

None of this shows that a parent action or any product caused development to change — and we won't pretend otherwise.

What the record shows

Some skills hand off. Others stay for life.

A skill's window isn't just when it appears — it's when it's actively being built. Some skills are stepping stones that get replaced by something more advanced. Others, once gained, stay. The assessment knows the difference, and stops asking about what a child has already moved past.

Babbling hands off to first words.

Babbling builds early — half of children by 5 months. As children move toward first words, the assessment shifts its attention from babbling to early speech.

0%25%50%75%100%0612182430child age, monthsBabbling →First words

Share of children showing each skill, by age · Kinedu record, 2020–2023 · shaded band ±1 SD across the skill's milestones

Walking arrives, and stays.

Half of children by 19 months, three quarters by 29 — and once gained, never handed off. The wide range is real: children reach it at genuinely different ages.

0%25%50%75%100%1218243036child age, monthsWalking

Share of children showing the skill, by age · Kinedu record, 2020–2023

What it watches

408 milestones. 48 skills. Four areas.

Every question is a concrete, observable behavior — written so anyone who cares for a child can answer just by watching. The expertise lives in the instrument, so it doesn't have to live in you.

👣
Physical
176 milestones

Motor development is the most granular to observe — from head control to jumping on one foot.

🧩
Cognitive
94 milestones

Cause and effect, problem solving, pretend play — thinking you can see.

💬
Language
79 milestones

From babbling to first words to sentences — and everything that hands off in between.

🙂
Social-emotional
59 milestones

Attachment, imitation, playing alongside and then with other children.

Our honesty rules

What this science refuses to do.

Every rule below exists because of a real mistake we made and fixed. That's what it means to be built on evidence.

It never diagnoses. It describes where a child is today and what usually comes next — developmental surveillance, not a medical label.

It celebrates upward, never alarms downward. A mistaken compliment doesn't hurt a family. A mistaken alarm does.

It shows no number it can't sustain. When eight answers aren't enough to place an area, it says “more answers needed” — never a scary zero.

Born early? It says so. Ages are corrected for prematurity using the American Academy of Pediatrics equation — and the reading declares it, every time.

It compares against a fixed norm. Your child is never ranked against other users — only placed on a frozen population reference, like a pediatrician's growth chart.

It won't promise what it can't measure. It doesn't score your parenting, and it doesn't pretend one snapshot defines a child. Development keeps unfolding.

Transparency

Don't take our word for it. Read the work.

The external research report and the Parent Guide are public, in three languages. Some periods are especially receptive — the receptivity narrows, it does not close. The full story, with methods and limits, is here:

Research report (PDF) → Parent Guide (PDF) →

Coupling and Differentiation in Early Development · July 2026 · Kinedu supports developmental surveillance. It is not a diagnosis.

Start science-based parenting today.

Everything in Kinedu is built on real research. See the difference it makes.

Start your free trial